To read about the newer RGB and data cameras see Digital Cinematography.

There is also a new pdf document available discussing the full 'Scene-to-Screen' requirements for DI, including HD capture. The document is available for those interested by using the 'Technical Papers' link to the left.

There are a plethora of HD cameras available from a range of manufacturers, including Sony, Panasonic, JVC, Thomson, etc., and they can all produce very good images, if used correctly.

The thing to remember is that the better the quality of the camera, the better the quality of the end result, especially if you are going to 'push' the image a lot in post-production. The smaller, heavily compressed HD cameras can produce good looking images, but the image will quickly fall apart in post-production, especially when graded heavily.

For this reason the information within this page will focus on the Sony f900 and Panasonic VariCam range of cameras, as they are the best 'standard' HD cameras presently available.

If using in HDcam mode (recording to internal tapes, not to an external vtr or disc recorder) the difference between the two is very small due to the compression Sony uses to get the 1920x1080 image onto the HDcam tape vs. Panasonic's 1280x720 HD format.

With the Sony HDcam format the full 1920 pixel image is reduced to 1440 pixels and compressed 8 to 1 via pre-filtering (frequency reduction) and reduced to an 8 bit signal. This is uncompressed back to 1920 pixels on playback. Therefore, the resultant image is similar to the Panasonic 1280x720 image, which also uses compression, but to a lesser extent.

At Digital Praxis we have used both with very good results, but always through the use of some form of 'Cine Gamma' mode (see .Digital Praxis Gamma Curves for more info.

This applies an OCN (Original Camera Negative) characteristic (Low contrast, high dynamic range) to the image on capture. Through such 'Cine Gamma' tools both cameras can be set-up to give good dynamic range on capture, either through the use of pre-set functions or via user definable gamma curves, allowing for great grading flexibility online.

Note: applied compression only affects the image detail, not dynamic range, and is not really a problem unless you intend to do keying or heavy secondary colour correction. The thing to avoid is attempting to 'grade' on-set.

Images captured via either camera can be used for offline editorial work, but both need to be 'reduced' for offline use. On thing to watch out for is that Avid fails to load with non-continuous timecode, which both cameras can cause if you don't rewind slightly between takes to ensure continuous timecode (don't use time of day timecode!).

The biggest issue is to make sure you map the linear data into log space during the grade. This will generate the best possible image as it provides a full range density image for printing.

Most HD projects tend to perform post work linear, and when recorded to film & printed there is not enough dynamic range in the original HD liner image to 'fill' the film log dynamic range. This results in lifted blacks and grey whites, especially if when printed the printer lights used are not 'on-aim' as this exposes further the limited dynamic range available. Technically, it prevents the film print toe & shoulder characteristics from doing their thing, which is a huge part of what 'film' is about. This is why all the linear HD projects Digital Praxis gets involved in have been mapped into log space as part of the DI process. This creates a full dynamic range negative that can be printed with wild abandon!

If you can manage this approach the results, regardless of the camera used, can be spectacular.

The following image, taken from the short feature 'Daddy's Boy', was shot with a Sony f900 in HDcam mode with Cannon zoom lenses (a typical 'video' package) and output to film in anamorphic 2.35:1. It's worth remembering that in HDcam mode the Sony records 1440 pixels (reduced from 1920), 8 bit pixel data (reduced from 10 bit) and reduces colour and luminance information via pre-filtering with a relative final compression ratio of 8:1. If Prime lenses had been used the results would have been even better, but as it was the results were amazing when screened (see story in film projects section).

This image shows the original HD linear image as captured with extended dynamic range. Notice no crushed blacks (in the hair for example) and no clipped whites (reflection in the picture above fireplace).

This shows the image mapped to Log space, with no LUTs applied. It looks a lot like a positive 'negative' (Interpositive), if you see what I mean.

This image is the Log image with LUTs applied, or exactly as a 25 across on-aim print looks. Notice again the lack of crushed blacks or clipped whites as the toe and shoulder regions of the D-LogE curve contained within the print film and/or output LUT are doing their thing!

<
From here we can print up and down, revealing more detail in the blacks or whites as required. Here the image is printed lighter to reveal more black detail, without the blacks turning grey.

For more in-depth information please review the 'Technical Papers' within the links at left of this page.